HEALTH EXPENDITURES 2016 edition BY DISEASES AND ...

S TAT I S T I C A L W O R K I N G P A P E R S
S TAT I S T I C A L W O R K I N G P A P E R S
M ain title
HEDIC HEALTH EXPENDITURES BY
DISEASES AND CONDITIONS 2016 edition
Europe Direct is a service to help you find answers to your questions about the European Union.
Freephone number (*):
00 800 6 7 8 9 10 11
(*) The information given is free, as are most calls (though some operators, phone boxes or hotels may charge you).
More information on the European Union is available on the Internet (http://europa.eu). Luxembourg: Publications Office of the European Union, 2016 ISBN 978-92-79-58542-5 ISSN 2315-0807 doi: 10.2785/434142 Cat. No: KS-TC-16-008-EN-N © European Union, 2016 Reproduction is authorised provided the source is acknowledged. For more information, please consult: http://ec.europa.eu/eurostat/about/our-partners/copyright Copyright for the photograph of the cover: ©Shutterstock. For reproduction or use of this photo, permission must be sought directly from the copyright holder. The information and views set out in this publication are those of the author(s) and do not necessarily reflect the official opinion of the European Union. Neither the European Union institutions and bodies nor any person acting on their behalf may be held responsible for the use which may be made of the information contained therein.
Acknowledgement This Working Paper presents the summarised results of the project on Health Expenditures by
Diseases and Conditions (HEDIC), carried out by colleagues in a consortium of institutions which
participated formally in the contract with Eurostat for this research. These institutions are listed in
Annex 9.
In research of this nature, involving the provision of data by a very wide range of institutions, a large
number of other people made important contributions to the work. In particular, the project team
would like to thank colleagues from the following organisations who attended workshops, contributed
to the methodological discussions, commented on drafts, and generally gave us the benefit of their
extensive knowledge of disease accounts: DG SANTE, the European Centre for Disease Control,
Eurostat, the Hungarian National Health Insurance Fund, OECD, the Polish National Health Fund,
and WHO. The team would like to acknowledge the contribution of colleagues from the Estonian
National Institute for Health Development, who were not members of the HEDIC consortium, but
nevertheless have attended workshops, and provided data for the HEDIC Pilot Data Set. It would
also like to thank colleagues in all of the agencies who provided data for the data inventory
questionnaire, sent to all EU and EFTA Member States at an early stage of the research.
The project team would also like to thank colleagues in the BASYS office for all their work to analyse
data, draft reports and assist with the logistics of the project. Finally, we are grateful to colleagues in
Socialstyrelsen Sweden, CBS Netherlands, and HCSO, Hungary, for organising the workshops
which took place during the project in Stockholm, The Hague, and Budapest.
4 Health expenditures by diseases and conditions
Contents
Contents
What data do we need to compile HEDIC? ................................................................................. 8
How does health expenditure vary by age and sex, over time and between countries?....... 8
How does health expenditure by disease vary, over time and between countries? .............. 9
Does health system design affect levels of expenditure by disease, age and sex? .............. 9
Can HEDIC data improve the international comparability of SHA? ......................................... 9
What is needed to incorporate HEDIC data in routine data collections? .............................. 10
How can we build on the work of HEDIC? ................................................................................ 10
2 Introduction ........................................................................................................... 11
Outline of the report ................................................................................................................... 13
Chapter summary ....................................................................................................................... 14
Health expenditure accounts as starting framework .............................................................. 16
Disease specific data.................................................................................................................. 17
Risk profiles ................................................................................................................................ 19
Chapter summary ....................................................................................................................... 21
Demographic structure .............................................................................................................. 23
Sex-specific issues ..................................................................................................................... 26
Grouping of diseases ................................................................................................................. 31
Contents
Variation of expenditures by disease between 2012 and 2013............................................... 33
Circulatory diseases ................................................................................................................... 34
Dimensions of health systems considered in SHA ................................................................. 41
Inpatient care............................................................................................................................... 41
Dimensions of the HEDIC statistical system............................................................................ 51
Summary of progress made in estimating expenditure by disease ...................................... 53
Practical considerations in compiling HEDIC .......................................................................... 53
Productivity loss ......................................................................................................................... 54
Incorporation into regular SHA data collection ....................................................................... 57
Chapter summary ....................................................................................................................... 58
Introduction This is the final report of the Eurostat-funded project, Health Expenditures by Diseases and
Conditions, HEDIC. In November 2013, Eurostat commissioned a 30-month programme of research
from a consortium consisting of representatives from 16 European Member States (MS). The primary
aim of the research is to develop further the methodology for the consumer health interface under the
System of Health Accounts (SHA), and hence to provide more detailed information on health care
expenditure in relation to its uses and beneficiaries, as a contribution to the public health statistics
available for monitoring EU health.
HEDIC and SHA The consumer health interface is of particular interest to the study of the relationship between the
consumption of health care goods and services and the associated health enhancement of the
population. Although health is only partly determined by the consumption of health care, the
breakdown of health care expenditures by characteristics of beneficiaries helps to improve the
understanding of the observed distribution in overall health spending. Health differences among
individuals and population groups are apparent along many dimensions, including age, gender,
socioeconomic status and geographic area. Age and gender are demographic characteristics of
beneficiaries that form an intrinsic epidemiological part of identifying and measuring the utilisation of
health care goods and services by type of disease.
Historically, the HEDIC project builds on work on cost of illness (COI) which began in the 1960’s with
Rice’s work on the economic burden of illness in the US economy, in which direct and indirect costs
of illness were estimated. HEDIC is different from COI. One reason is the exclusion of indirect costs.
Indirect costs in the estimation of COI, or productivity loss, measure the loss in earnings as a result
of death, illness or time spent undergoing treatment for the population as a whole. HEDIC is focusing
on direct costs by using the framework of SHA, which offers the possibility of developing consistent
expenditure by disease accounts across countries. Employing consistent methodology and data
sources can ensure that expenditures for various diseases can be compared and that the sum of
expenditures for all diseases equals the estimate of current health expenditure.
Recognising the importance of further developing the methodology for health expenditures by
disease to develop public health statistics for monitoring EU health, Eurostat issued an Invitation to
Tender. A consortium of representatives from 16 EU MS was commissioned, and has carried out 5
main tasks to deliver the work:
Task 1 involved compiling a data inventory, which describes the data available on
expenditure by disease, age and gender, in EU Member States and EEA/EFTA countries.
Members of the HEDIC consortium completed a questionnaire describing the availability of
data on health expenditure by diseases and conditions in their countries.
1 Executive Summary
1 Executive Summary
Task 2 comprised the preparation of a HEDIC Manual, which sets out the guidelines for the
compilation of expenditure by disease data.
Task 3 refers to the pilot projects which were carried out by members of the consortium, in
order to test the proposed methodology for compiling expenditure by disease data.
Task 4 involved detailed analysis of the pilot project data.
Task 5 refers to the three workshops which took place in Stockholm (June 2014), The
Hague (March 2015) and Budapest (April 2016).
What data do we need to compile HEDIC? HEDIC requires both macro and micro data, as well as metadata. At the macro level, it aims to
allocate SHA current health expenditure by disease, age and gender. Because different countries
compile SHA from different data sources, the HEDIC methodology is flexible, to take this into
account. In cases of multi-morbidity, the current convention is to attribute expenditure to the primary
diagnosis. Where countries used a different method they are asked to specify this in their metadata.
Where possible, countries extract micro data for distributing expenditure by disease age and gender
directly from provider or financing data sources. Where data on expenditure by disease was lacking,
countries were asked to estimate this using utilisation data and data sources for making unit costs
estimates. For pharmaceutical expenditure, countries were invited to use a database of mapped
ATC-ICD codes to allocate expenditure.
A manual describing the HEDIC methodology and a data inventory questionnaire was sent to
representatives of National Statistical Authorities, Ministries of Health, or research consultancies, in
all EU and EFTA countries. Analysis of their responses enabled a HEDIC Pilot Data Set (HPDS) to
be specified. Members of the consortium submitted data using this framework.
How does health expenditure vary by age and sex, over time and between countries? Age and gender-related health expenditure profiles are extensively used in forecasting models of
health expenditure, and are important for international comparisons, because demographic
structures vary greatly among countries. Chapter 3 discusses:
1. variation in health expenditures by age among countries in a given year
2. the change in expenditure profiles, particularly the steepening of HE by age, over time
3. sex-specific issues of health expenditure by age.
It presents data on expenditure by age as a percentage of current health expenditure for nine
countries in 2013; and compares the expenditure profile by age for five countries, in 2012 and 2013.
In order to compare health expenditure between men and women, it is very important to separate the
cost for pregnancy and reproduction from other costs, where it is common to attribute these costs to
the mother. The same applies to sex-specific diseases such as ovarian and prostate cancer. Health
expenditures by age for women’s health by ICD-10, for conditions related to pregnancy, are
presented for pregnancy, childbirth and the puerperium, in 2013.
The chapter concludes by discussing the use of age-related expenditure profiles for expenditure
forecasts.
How does health expenditure by disease vary, over time and between countries? HEDIC shows major variations in health spending by disease in contrast to former international
comparisons. This may be because those studies compared a limited set of providers of Western
European countries, but excluded countries in Eastern Europe. Furthermore, differences might be
partly explained by differences in exclusion/inclusion of specific functions and providers included in
the data. Chapter 4 examines these variations, before looking more closely at three important
disease categories.
Data on health expenditure by disease as a percentage of allocated current health expenditure are
presented for 2013, for eleven countries; and the deviation of the growth rate of health expenditure
by disease from national average between 2012 and 2013 is compared for six countries. For
circulatory disease, neoplasms, and mental disorders, data submitted for the HPDS are compared
with those presented in other published studies, and the differences in levels of expenditure are
commented on.
Does health system design affect levels of expenditure by disease, age and sex? Data submitted for the HPDS show major variations in the share of expenditure on pharmaceuticals
and inpatient care by diseases among countries. The challenge is to distinguish between absolute
differences in expenditure on diseases among countries, and those differences which are artefacts of
the data available for measuring potential differences. In Chapter 5 we focus on the two areas of
pharmaceuticals and inpatient care, in discussing the possible impact of health system design on the
distribution of expenditure by disease.
2013 data for ten countries in the HEDIC consortium are presented: on health expenditure on
inpatient care by disease as a percentage of total inpatient care; on volume of inpatient care by
disease as a percentage of total inpatient volume; and on relative unit costs of inpatient care by
disease.
Many countries do not have accurate outpatient medication data by disease, but all countries classify
pharmaceutical expenditures by the Anatomical-Therapeutic-Chemical Classification System (ATC).
Various methods for mapping ATC data to ICD data are proposed. Data on the distribution of
pharmaceutical expenditure, on the volume of this expenditure, and on unit costs of pharmaceutical
expenditure, by disease, are presented.
Can HEDIC data improve the international comparability of SHA? As discussed above, HEDIC adds additional information to health expenditure comparisons. Chapter
6 assesses whether HEDIC would also improve the compilation and comparison of the three core
tables of SHA.
Including information on expenditure by disease, age and gender in SHA will improve the
international comparability of SHA, by improving the three core tables of SHA, for the following
reasons:
Countries will interrogate their data sources more thoroughly in order to compile these
additional dimensions, thereby leading to improvements in the quality of the SHA
1 Executive Summary
compilations, and, in some cases, an increase in the number of data sources used.
We will gain a better understanding of different levels of spending by function, if we know the
age profile of the users of services in different countries. For example there are some
important inter-country differences in the age of users of long-term care.
Many of the countries participating in the HEDIC project agree that their search for the data needed
to compile HEDIC, has helped them to understand the structures and development of health
expenditures within their countries, and as compared to other countries.
HEDIC can help the comparison of existing data collected by Eurostat and OECD by adding more
flesh to the bones. For example, it is easier to understand the differences in SHA pharmaceutical
expenditures or inpatient care if we have information on distribution by age, sex and disease in
addition to the core tables of SHA. Furthermore, HEDIC allows the standardisation of health
expenditure.
What is needed to incorporate HEDIC data in routine data collections? The HEDIC project has attempted to assess the effort needed to incorporate routine collection of
data on health expenditure by disease, age and sex, in the European Statistical System. While the
HEDIC study has demonstrated the general feasibility of collecting data on expenditure by age, sex
and disease, any decision to collect such data routinely must also take into account the resources
available for doing so within countries, and countries’ current intentions and plans for continuing to
work in this area. Countries actively participating in the HEDIC project were asked to describe their
current plans for work on disease accounts, and to estimate the resources they would need in terms
of appropriately qualified person(s) working in the organisation with principle responsibility for
compiling disease accounts. Twelve of the fourteen countries supplying data to HEDIC stated that
they plan to continue work on disease accounts. Estimates of the resources needed for regular
compilation of disease accounts in the national organisation responsible for this work ranged from
0.25 to 1.9 Full Time Equivalent staff per annum. This range reflects the current state of development
of disease accounts in different countries.
Other important considerations relate to the possible need for formal legal arrangements to facilitate
inter-institutional exchange of information between institutions within countries; and the importance of
maintaining the momentum and expertise developed during the HEDIC project
How can we build on the work of HEDIC? HEDIC has made considerable progress in estimating expenditure by disease in European countries:
for more countries than hitherto, HEDIC can show costs of diseases by ICD chapters; it has compiled
health expenditure profiles by age and sex which allow better projections of future health
expenditure; and health expenditure values of inpatient care and pharmaceuticals have been split in
to volumes and prices. A close link to non-expenditure statistics has been established.
In the final chapter it is recommended that further work be carried out to assess the practical aspects
of incorporating HEDIC in the routine health expenditure data collection of Eurostat, in order to
improve the analytical capacity and international comparability of SHA. It is argued that it will be
desirable to do this because we will be better placed to understand health-specific cost-drivers, and
hence contribute to the debate on the disease-specific interventions of health systems, if we have
information on expenditure by disease and age. For example, combining information on trends in
pharmaceutical costs by disease chapter, and demographic information, will help Member States to
understand how and why their country differs from the EU average, and to separate local from
international trends.
2 Introduction
Background Studies of health expenditures by disease and conditions have a long tradition. They are closely
linked to Cost of Illness (COI) studies which measure the economic burden of a disease or diseases.
COI studies examine the allocation of resources from different perspectives. Who is affected? On
whose behalf are decisions made? Depending on the perspective taken they may measure cost to
society as a whole, health financing schemes, health care providers, households and/or different
population groups. Disparities in health care spending are found within populations along many
different social dimensions, all of which may be of policy and analytical interest. Dimensions of
particular interest include the type of disease or health care condition, age, sex, geographic area and
socioeconomic status.
Since at least the mid-1960s variations in health care expenditures within national populations have
been analysed. Rice (1967) made the first attempts to measure the variations in spending by
disease, age and sex in the United States. She estimated the national economic burden of all illness
in the United States for 1963, from a societal perspective. Her analysis focused on two main types of
costs - those of health care resources (direct costs), and those of productivity losses resulting from
illness (indirect costs). Rice also noted another cost component – the “intangible or psychic costs” of
disease such as pain and grief. The methodology employed became the accepted general
framework for COI studies, and is still used in many studies today.
In general, direct costs refer to the value of resources used as a result of disease. With reference to
SHA boundaries used in the System of Health Accounts (SHA), they can be divided further into direct
health costs and direct social costs1. Direct health costs refer to those costs that are within the
boundary of health care expenditure as defined by SHA 2011. Social costs refer to expenditures
associated with “social care” as defined by SHA 2011. These are goods and services indirectly
related to the provision of health care which are outside the health care boundaries.
Disease accounts compiled from a societal perspective provide a comprehensive picture of
population health relative to health care spending. These accounts provide a lot of useful information,
and should be viewed as one piece of information, or one input, into the decision-making process. It
can be argued that policy makers should not make decisions based solely on the results of COI
studies. COI studies do not purport to focus on health interventions and their effectiveness. That is
left to the field of economic evaluation. COI studies can, however, provide very valuable information
for policy makers. In particular, the results from such studies can be used as an input into further
types of analyses such as cost-benefit or cost-effectiveness analysis.
As discussed in SHA (2011, p. 227), although health accounts expenditures were already applied to
1 Social costs, as defined in SHA, would include those components of long-term care not directly related to health or additional social
care, for which payment has been made.
2 Introduction
2 Introduction
disease-specific studies from the early 1990s, the use of standard classifications of expenditure, and
in particular SHA, has improved the usefulness of such studies. Various projects since 2000 have
assessed the feasibility of analysing health spending by beneficiary characteristics. Both Eurostat
and OECD have jointly collaborated on projects to develop a set of guidelines, based on the
pioneering work by RIVM, for the distribution of spending by disease, age and gender, which have
subsequently been tested in various European MS.
Koopmanschap (1998) gives a detailed summary of some of the potential uses of COI studies:
Providing information on the burden of specific diseases;
Estimating disease costs covering the entire classification of diseases, enabling mutual
comparison of disease costs and putting these in perspective;
Prioritizing diseases or topics for future economic evaluation (i.e. by combining COI data
with other information such as information about effectiveness of treatment);
Incorporating COI results in cost-effectiveness analysis, e.g. as a cost estimate of current
treatment which can be compared with the program studied;
Clarifying the most important cost components of treating specific diseases; and
Explaining recent trends in costs and/or projecting future disease costs, based on
demographic, epidemiological and technological change (i.e. when COI data are used as a
component of scenario-analysis).
One of the main benefits of using a comprehensive health accounting approach is that all
expenditures are allocated to different disease groups in a mutually exclusive manner, which is
important in light of the co-morbidities of chronic diseases (see chapter 3). This avoids the issue of
double counting which can occur in studies focusing on selected diseases; if the same transaction
gets counted in two different studies (i.e. can be linked to two different diseases).
Aims of HEDIC The project ‘Health Expenditures by Diseases and Conditions (HEDIC)’ contributes to Eurostat’s
work on “Public health statistics for monitoring EU health”, which aims to increase the use of official
public health data at EU level. It will provide important information on burden of diseases by linking
health expenditure data with patient characteristics. This builds specifically on three projects carried
out over the last fifteen years. The first was a systematic attempt to arrive at a breakdown by patient
characteristics of health care expenditure data classified by function, age and gender for the years
1999 or 2000 (IGSS, CEPS 2003). The second project focused on breaking down health
expenditures by age, sex and diseases for all EU Member States and EEA/EFTA countries, including
a suggestion of a shortlist of diseases/disease categories for selected ICD chapters (BASYS, CEPS,
IGSS 2006). In a further project, the OECD (see OECD 2012), collected additional information about
expenditures by disease for several EU Member States. HEDIC builds on the experiences of these
earlier projects.
HEDIC complements several other strands of work in the EC. A Task Force on Morbidity Statistics,
established in 2011, is overseeing Eurostat’s work to develop diagnosis-based morbidity statistics, in
order to fill an important gap in the information available on the health status of the EU population.
This information is crucial for the development of public health indicators at the EU level. From 2005
to 2011, 16 MS participated in pilot studies on diagnosis-specific morbidity statistics. In 2014, the
Task Force presented a report on the in-depth analysis of these pilot studies, and made
methodological recommendations with regard to sources and best estimates (Eurostat 2014). DG
ECFIN’s work on long-term age-related expenditure projections, which aims to provide insights into
the economic impact of ageing, includes work to project health care expenditure. In developing its
health expenditure projections it uses age-gender-specific expenditure profiles supplied by EU MS
(European Commission 2014). DG SANTE work to develop and maintain the European Core Health
2 Introduction
Indicators will also benefit from and inform the collection of HEDIC data, where these indicators
require accurate and internationally comparable data on expenditure by disease2.
Attributing health care expenditures to diseases and conditions, and demographic characteristics of
age and sex, provides basic information on current resource allocations in the health care system
related to the morbidity of the population. This HEDIC information can inform current discussions
concerning ageing populations and changing disease patterns, by analysing time trends, identifying
the drivers of health care spending, and providing an input into modelling of future health care
expenditures (European Commission 2014). Furthermore, the linking of health expenditures to
measures of utilization (e.g. hospital discharges by disease), prices (e.g. DRGs), and outcomes (e.g.
myocardial infarctions) can provide a useful input in the analysis of health expenditure development
and in monitoring the performance of health care systems.
Work carried out to deliver HEDIC Eurostat awarded the HEDIC contract to a consortium consisting of representatives from National
Statistical Authorities, Ministries of Health, Social Insurance organisations, and research
consultancies, from 16 EU Member States (these are listed in Annex 8.3).
Five main tasks were delivered:
Task 1 involved compiling a data inventory, which describes the data available on expenditure by
disease, age and gender, in EU Member States and EEA/EFTA countries. Members of the HEDIC
consortium completed a questionnaire describing the availability of data on health expenditure by
diseases and conditions in their countries.
Task 2 comprised the preparation of a HEDIC Manual, which sets out the guidelines for the
compilation of expenditure by disease data.
Task 3 refers to the pilot projects which were carried out by members of the consortium, in order to
test the proposed methodology for compiling expenditure by disease data.
Task 4 involved detailed analysis of the pilot project data.
Task 5 refers to the three workshops which took place in Stockholm (June 2014), The Hague (March
2015) and Budapest (April 2016).
Outline of the report Chapter 2 discusses HEDIC data requirements, and reports on HEDIC data availability as reported in
the data inventory questionnaire sent to all European MS.
This study has analysed pilot HEDIC data supplied by members of the HEDIC consortium using four
groups of hypotheses, and these are discussed in Chapters 3 to 6.The first group deals with
demographic issues related to presentation of health expenditure by age and sex, this being an
important issue in ageing European societies. The second group of hypotheses focuses on disease
related questions. As HEDIC comprises the whole landscape of morbidity, it is possible to compare
the results of HEDIC with cost-of-illness studies for specific diseases. We compare HEDIC estimates
with European cost-of-illness studies for major ICD chapters. The third group of hypotheses relates
to aspects of health system design. We investigate whether the organisation of a health system has
an impact on levels of expenditure on different diseases, and whether those differences are artefacts
of the data available for measuring potential differences. Finally, the fourth group focusses on the
2 http://ec.europa.eu/health/indicators/indicators/index_en.htm (Accessed 14/03/16).
statistical measurement of health expenditures, and in particular, asks if HEDIC can improve the
comparability of compilations of SHA. Chapter 6 discusses what will need to be done to develop
HEDIC further, and Chapter 7 concludes the report, and makes recommendations for incorporation
of HEDIC in the European Statistical System. Chapters 3 to 6 conclude with an ‘Outlook’ section
which discusses further issues relevant for the future compilation of HEDIC.
Chapter summary The HEDIC project is a continuation of work on cost of illness which began in the 1960’s
with the work of Rice on variation in health spending by disease, age and sex in the USA.
Information from COI studies can be applied in a variety of ways, including assessing the
burden of specific diseases, comparing disease costs across the entire spectrum of
diseases, in cost-effectiveness analysis, and in explaining recent trends and forecasting
future health care costs.
From the early 1990s, the use of standard classifications of expenditure, and in particular
SHA, has improved the usefulness of disease-specific studies of health expenditure.
HEDIC builds directly on three earlier projects funded by Eurostat since 2000, examining the
feasibility of routine data collection to examine the link between health expenditure and
patient characteristics. It complements Eurostat’s work to develop morbidity statistics, DG
ECFIN work on ageing, and DG SANTE work on ECHI.
Recognising the importance of developing the methodology for costing illness in order to
develop public health statistics for monitoring EU health, Eurostat issued an Invitation to
Tender for work to: make an inventory of the data sources available for costing illness in
European MS; write a manual for the construction of HEDIC; and to collect and analyse a
HEDIC Pilot Data Set from countries able and willing to supply this data.
Eurostat awarded the HEDIC contract to a consortium consisting of representatives from
National Statistical Authorities, Ministries of Health, Social Insurance organisations, and
research consultancies, from 16 EU Member States. The work was carried out between
November 2013 and May 2016.
References BASYS, CEPS and IGSS (2006), “Feasibility study of health expenditures by patient characteristics”,
Eurostat Grant: 2004 35100 018, Final report, BASYS, Augsburg.
European Commission (2014), The 2015 Ageing Report: Underlying Assumptions and Projection
Methodologies, Brussels, European Economy 8/2014.
Eurostat (2001), Handbook on price and volume measures in national accounts, Luxembourg,
http://epp.eurostat.ec.europa.eu/portal/page/portal/product_details/publication?p_product_code=KS-
41-01-543.
Eurostat (2014), Morbidity Statistics in the EU: Report on pilot studies, Eurostat: Luxembourg.
Henderson J (2012), Cost of Illness, Indirect Costs, and Mental Health. Health Accounts Experts
Meeting, OECD Paris, 13- 14 February 2012.
IGSS, CEPS (2003), Age and gender-specific functional health accounts: A pilot study of the
application of age and gender-specific functional health accounts in the European Union. Final report
IGSS/CEPS, November 2003, Luxembourg.
IRDES, BASYS (2007), “Tools for data collection on health care statistics”, Eurostat Grant:
35100.2005.012-2005.825, Final Report, IRDES, Paris.
Koopmanschap, M.A. (1998), Cost-of-illness studies, Useful for health policy? Pharmacoeconomics.
14(2):143-8.
OECD, Eurostat, WHO (2011), A System of Health Accounts 2011, OECD Publishing, Paris.
OECD (2012), Expenditures by Disease under the SHA Framework 2012 Project, Draft Interim
Report, Health Accounts Experts Meeting, OECD Paris, 13- 14 February 2012.
Rice DP (1967), Estimating the cost of illness, Am J Public Health, 57(3):424–440.
Shiell, A. and Gerard K. Donaldson (1987), Cost of illness studies: an aid to decision -making?
Health Policy, 8:317-23.
Wiseman, V., and G. Mooney (1998), Burden of illness estimates for priority setting: a debate
revisited, Health Policy, 43:243-51.
Health expenditure accounts as starting framework HEDIC is using available information both from macro statistics and from micro data. From the
macro-perspective, which is recommended by SHA 2011, the challenge of HEDIC might be
described as follows. Suppose the annual health expenditures by disease and conditions for the
population of a country is described by the vector, f, for 1,…., k-type conditions. Under the
assumption that the total value of f equals the current health expenditures, g, compiled by SHA, one
can derive the HEDIC vector f from SHA by premultiplication of the vector, g, with a coefficient matrix
Φ (expenditure items x health conditions).
(1) =
This coefficient matrix Φ with the dimension (k x n) consists of k rows for health conditions and n
columns of types of health expenditures by activities (e.g. expenditures by health care functions).
Figure 1 shows the structure of this matrix, the case of Germany by 44 ICD groups and 15 health
expenditure items.
The pattern of coefficient matrix Φ depends very much on the specialization of health care providers:
whether their activities are directed to specific diseases, e.g. dental care, or are general, like general
medicine, general hospital care, and pharmaceuticals.
An important issue in computing the coefficient matrix Φ is that in many cases multiple issues (multi-
morbidity) underlie the consumption of a unit of health care. There are three potential options for
dealing with this issue (see SHA 2011, OECD 2008, 2013):
1. to classify expenditures according to the primary diagnosis;
2. to equally pro-rata the expenditures over the applicable diagnoses;
3. to distribute expenditures across the applicable diagnoses using disease-specific weights
that reflect the relative resource intensity involved.
3 Methodology and Data Requirements
Figure 1: Example: structure of a matrix Φ (probability map) in the case of Germany 2008
Source: BASYS.
Although the third is conceptually the most ideal and provides a clear link between disease and total
spending, in practice the data requirements to support such an approach are immense and in most
countries will not currently be met. The second option is more feasible, but in many situations the
available data will only have recorded the primary diagnosis, and not all co-morbidities. Given this, it
is generally agreed that the standard approach should follow the first option. That is, to classify
expenditures according to the primary diagnosis, except in those instances where the primary
diagnosis cannot be differentiated from other diagnoses in the available data, in which case the
expenditures should be pro-rated equally across all relevant conditions. Therefore, the HEDIC
methodology recommends that health care costs should all be attributed to the primary3 diagnosis, if
the hierarchy of diagnosis is known.
However, in many cases this hierarchy is not well known, or is open to question. For instance if the
diabetic condition of a patient causes kidney failure, should these costs be attributed to diabetes as
the underlying condition, or to kidney failure, as this is the condition which generates the use of
health care? An alternative to attribution to the primary diagnosis is a proportional division over all
diseases and conditions, preferably weighted, so outcomes are adjusted to the severity of a disease.
Countries should provide information on which method was/will be used in compiling expenditure by
disease.
Disease specific data The micro information for distribution of expenditures by HEDIC categories should be extracted
directly from financing or provider sources. From a technical point of view, this extraction should
include items of the core dimensions of SHA (health care functions, provisions, and financing) as well
as the HEDIC categories. Countries are free to collect these data on a very detailed level and to be
3 In practice, this might be the discharge diagnosis.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
flexible with regard to national and international reporting.
As with information collection on age and gender, the majority of EU countries collect diagnostic
information at specialist hospitals, general hospitals and outpatient facilities, but not at residential,
day treatment or primary health care facilities.
Utilisation and unit cost Utilisation unit and cost unit must correspond to each other. Utilisation is usually measured in
different units for different expenditure items, such as hospital days, number of patients treated,
number of procedures performed, contact time, etc. Information on the utilisation unit is collected in
the metadata.
Data on utilisation are important for three reasons:
1. Data on expenditure by disease may not be available, but the necessary information for their
estimation (e.g. utilisation data and data sources for making unit costs estimates) may be
available.
2. The breakdown of expenditure by the coefficient matrix Φ for each expenditure item into
volume and price, can help countries to understand better the changes in expenditures over
time.
3. The decomposition of the expenditures by disease across European countries can help to
explain the reasons for the variation (see Dunn et al 2013).
For the compilation of the matrix Φ it is useful to understand the utilization of health care activities
and prices. Each column of the matrix Φ can be decomposed further into the volumes and prices by
disease categories. The probability vector of the health activity j, denoted , is the product of the
diagonal matrix of disease prices4 (.)and the utilization vector, x.j divided by the total
expenditure for the activity j, the scalar . This leads to equation (2)
(2) . = (.) . ⁄
Often, one cannot observe the prices directly, only the volumes of utilization by diseases which are
gathered by non-expenditure statistics. Therefore, the disease based prices are derived from values
and volumes.
(3) . = (1 .) ⁄ (. )
The vector ., represents the vector (k × 1) of disease based prices. The first term on the right side
of equation (3), the diagonal matrix = (1 .) ⁄ shows the reciprocal of the volumes 1/. of health
activity j by all diseases and conditions. The second term, (. ), presents the expenditures for the
health activities by disease. This decomposition helps tracing back changes in HEDIC in volume and
prices, and in a further breakdown into prevalence rates and access to care.
Although comprehensive information about unit costs is routinely collected at the level of health care
providers (see Eurostat, OECD 2012), information about prices by disease groups is still limited.
Such prices are presented in chapter 0 for inpatient care and in chapter 0 for pharmaceuticals
4 See Bradley et al (2010) for discussion of disease based prices.
Risk profiles Of particular interest are deviations of expenditures across population groups. For comparisons
across health activities and population groups, we must standardize the expenditure ratios, volumes
and prices. Suppose the column vector provides the values of the distribution of health
expenditures across age groups. Divided by the respective numbers of individuals in each age
group, , we receive the average expenditure per age (/). Standardized by the average
expenditure over all age groups we receive a profile which varies around the mean.
(4) ∗ = (/) /(/)
The vector i in formula 4 presents the unit vector for summation over all age cohorts.
Metadata Metadata is “data about data”. It is descriptive information about a particular data set, object, or
resource, including how it is formatted, and when and by whom it was collected. For the national
purposes of construction and updating of HEDIC at a minimum, background information should
include the sources of data, data items, how data were validated (especially in the case of multiple
data sources), the reasoning behind the selection of the data used in the estimation, the procedures
applied to make the data usable, and more. Solid, comprehensive metadata facilitates an appropriate
interpretation and use of the health accounts results. For example, trends in health expenditures by
disease groups can be analysed better when there is knowledge about, say, changes in the
accounting system.
Hence, in the data inventory questionnaire (discussed below), and in subsequently supplying pilot
HEDIC data, countries were asked to provide metadata.
Data availability An inventory of data sources for compiling the HEDIC Pilot Dataset (HPDS) was prepared, based on
information supplied by representatives from Ministries of Health, Central Statistical Offices, National
Insurance Organisations and consultancy organisations, in EU MS and EFTA countries. A manual
describing the HEDIC methodology was developed in a series of workshops involving HEDIC
participants, and representatives from international organisations. The manual describes in detail
how to compile the HEDIC data set. The methodology is flexible, in that countries may use any of the
SHA dimensions of functions, finance and activity, as their starting point in compiling HEDIC.
A questionnaire requesting information on HEDIC data sources was sent by the project in March
2014, to those EU MS and EFTA countries participating actively or as observers in HEDIC, and in a
second round, to all other MS and EFTA countries not involved in the HEDIC project. In total, 19
countries supplied detailed information about the data sources available for compiling HEDIC. The
questionnaire requested information in six sections:
country contact details;
indirect cost;
COI studies involving the country responding to the questionnaire metadata.
It is important to note that the expenditure items selected for the breakdown by disease, age, and
gender, are likely to be different among MS, because data sources vary, reflecting differences in the
organisational structure and approach to financing of different countries’ health care systems. For
example, one MS might only report the breakdown for acute hospitals, another for all hospitals.
A data inventory was prepared based on the responses to the questionnaire described above. This
helped to assess the suitability of these sources for compiling a HEDIC Pilot Data Set (HPDS). The
content of the HPDS was agreed at the second HEDIC workshop in Den Haag, based on the
following criteria:
Need for harmonisation of the assumptions, procedures and compilation rules including
definitions and classifications, with the statistical requirements;
Possibility of linking data to other sources (unique identifier/key variables or possibility of
using probabilistic linking);
Availability of metadata.
Assessing data available for compiling HEDIC To date fourteen countries have delivered the HPDS, with some differences in the level of detail (e.g.
of age groups) and years covered. Table 1 shows the variation of reporting dimensions of the data
sets provided by countries, for Allocated Current Health Expenditure (ACHE)5. The reasons for these
variations are manifold, and reflect legal, technical and financial aspects of compiling HEDIC.
Protection of personal health data has priority in all European countries. It is not permitted to process
these data for non-medical reasons. Under certain conditions however, statistical analysis in
anonymised form or at least with secure pseudonymisation is possible. In the case of HEDIC,
anonymised data for population groups are sufficient for the compilation.
Other reasons for differences in the coverage of the HPDS include:
Variation in national standards used for age groups used to compile administrative statistics;
Absence of diagnostic coding for some disease categories;
Lack of diagnostic coding for some types of hospital case;
The need for formal legal arrangements to be in place for inter-institutional exchange of
data;
Lack of resources currently available for processing large volumes of data.
5 The term “Allocated Current Health Expenditure” is used because, despite using a top-down approach some expenditures are not
allocated to disease because no disease-specific information for this expenditure item is available. For example, not all hospital cases are classified by disease.
Table 1: Characteristics of HEDIC Pilot Data Set supplied by countries
Items \ countries BG CZ DE EE EL LV LT LU HU NL AT SI FI SE
1. Expenditure
CHE x x x x x x x
CHE Public x x x x x x x x() x x
CHE Private x x x x x x x()
2. Breakdown by age groups and gender
by age (number of groups) 3 21 21 21 18 21 21 21 19 21 21 21
by gender (3 items) 3 2 3 2 2 2 3 3 2 2 2 2
3. Table by ICD 10 chapters
by ICD 10 chapters (number) 20 22 22 22 22 22 22 22 22 18 20 22 21 22
and by age (number of groups) 3 2 21 21 21 21 21
and by gender x x x x x x x x x x
and by age x gender x x x x x x x
4. Price and volume measures by disease
Inpatient (expenditure) x x x x x x x x x x() x x x x
Inpatient (unit cost) x x x x x x x x x _ x x x
Inpatient (volume) x x x x x x x x x _ x x x
Pharmaceuticals (expenditure) x(¹) x x x() x x() x x x
Pharmaceuticals (unit cost) x(¹) x x x() x x() x
Pharmaceuticals (volume) x(¹) x x x() x x() x
5. Years
2011 x x
2012 x x x x x x x x x x
2013 x x x x x x x x x x x x
2014 x x x
Primary care x (²) x x x
Long-term care x (³) x x
Physiotherapy x x x x x() x
OTC Market x x x x x x x x x
Public health (²) x x x
Administration (²) x x x x x
(¹) prescribed medicines reimbursed by health insurance only (²) included in CHE, but complete allocation to ICD chapters was not possible (³) inclusion of LTC in the current SHA is likely incomplete and will be improved in the near future () beside pharmaceuticals inclusion of other medical goods
() refers to hospitals
() only applies to the year 2012
() physiotherapy financed by Public Health Insurance Fund and Private Health Insurance Companies are included; physiotherapy financed by households is not included.
Chapter summary HEDIC uses information from macro statistics and from micro data, and recommends that
expenditure be attributed on the basis of primary diagnosis, in cases where the hierarchy of
diagnosis is known.
Microinformation for distribution of expenditures by HEDIC categories is extracted directly
from financing or provider sources, recognising that this is most likely to be available for
specialist hospitals, general hospitals and outpatient facilities, but not at residential, day
treatment or primary health care facilities.
Data on utilization and unit cost are collected for 3 reasons: to estimate expenditure where
cost by disease data is not directly available; to help to explain changes over time; and to
assist in explaining international differences in disease-related costs.
Countries are asked to supply metadata, for national purposes of constructing and updating
HEDIC, and to assist in interpreting international differences. This should include the
sources of data, data items, how data were validated (especially in the case of multiple data
sources), the reasoning behind the selection of the data used in the estimation, and the
procedures applied to make the data usable.
A data inventory questionnaire was completed by all participants in the HEDIC project and
three other MS. MS used an early draft of the HEDIC Manual to assist them in preparing
their responses. Compiling the data inventory also informed subsequent development of the
manual.
Responses to the questionnaire were used to assess the suitability of data sources for
compiling a HEDIC Pilot Data Set (HPDS).
Fourteen countries have delivered the HPDS, in varying levels of detail and for different
years.
References Bradley, R., Cardenas, E., Ginsburg, D., Rozental, L., Velez, F. (2010), Producing disease- based
price indexes, in: Monthly Labor Review, February 2010: 20 - 28.
Dunn A., Shapiro A.H., Liebman E. (2013), Geographic variation in commercial medical-care
expenditures: A framework for decomposing price and utilization, in: Journal of Health Economics 32
(2013): 1153-1165.
Eurostat (2001), Handbook on price and volume measures in national accounts, Luxembourg,
http://epp.eurostat.ec.europa.eu/portal/page/portal/product_details/publication?p_product_cod e=KS-
Luxembourg.
OECD (2008), Estimating Expenditure by Disease, Age and Gender under the System of Health
Accounts (SHA) Framework.
OECD, Eurostat, WHO (2011), A System of Health Accounts 2011, OECD Publishing, Paris, OECD
(2013), Extension of work on expenditure by disease, age and gender, EU Contribution Agreement
2011 53 01, December 2013, Health Division, www.oecd.org.
Demographic structure The need for health varies by women and men over the life-cycle. It is important to understand the
utilisation of health services over this cycle in light of the respective population age structure.
Expenditure profiles by age and sex summarise the individual expenditures at a given point in time.
In addition to the impact of age on the need for health care, several other factors affect the forms of
these profiles, such as the organisation of health care, and access to services in different age
cohorts.
Interest in the analysis of expenditures by age and sex has grown, with increasing attention being
given to the implications of population ageing for health care system organisation and health care
financing. Health expenditure profiles by age and sex are extensively used in forecasting models of
health expenditure (see Astolfi et al 2012, European Commission 2014). They are also important for
international comparisons, because demographic structures vary greatly among countries (see
Finkenstädt, Niehaus 2015). Below we discuss the following aspects of such expenditure profiles:
1. variation in current health expenditures by age among countries in a given year
2. the change in expenditure profiles, particularly the steepening of expenditure by age, over
time
3. sex-specific issues of health expenditure by age.
Our analysis of health expenditure profiles uses the following data from the HPDS submitted by
countries: Current health expenditures by age and male, by age and female, for both 2012 and 2013,
and the respective population data. Although the changes of profiles are rather small in such a short
period, it is important to understand these changes.
Variation of health expenditure by age across countries This subsection describes the variation of allocated current health expenditure by age among HEDIC
countries.
Following the methodology of HEDIC, all expenditures should be allocated by age. The distribution of
expenditure depends on the number of individuals in each age class. For example, one can expect a
higher share of expenditure in the class 85-89 in a country with more elderly people. In contrast one
would expect that countries with relatively fewer births spend less for the age class 0 (see Table 2).
In fact, Germany has the lowest birth rate and shows the lowest expenditure share for the age class
0. However, Sweden, with the highest birth ratio, shows an expenditure share of 2.1 percent, but
below Latvia, with a share of 2.3 percent despite a lower birth ratio. Clearly, other factors contribute
4 Health expenditure profiles by age and sex
4 Health expenditure profiles by age and gender
to this distribution such as availability of services for higher age groups.
(%)
(¹) 2012 instead of 2013 (:) not available
The variation in health expenditure by age among HEDIC countries is not only determined by the
structure of the population, but also by the risk profiles for the individuals depending on their age.
Risk profiles show the average variation in health expenditures for individuals. We compiled these
risk profiles dividing the health expenditures for each age class by the respective population and
standardized them by the mean. These standardized expenditure profiles per capita are presented in
Figure 2. Health expenditures per capita are divided by the average expenditures per capita of the
respective country. They show expenditure in a given age class as compared to the average.
For long-term expenditure projections in the field of health care and long-term care it is necessary to
decompose the figures of Table 2 into expenditure per capita and number of individuals (see
European Commission 2014).When interpreting the data in Table 2 and Figure 2, the proportion of
health expenditures without recorded age should be born in mind (e.g. in the case of Czech Republic
and Lithuania).
It is also important to note how differences in the organisation of care may affect these projections. In
the Netherlands, for example, relatively more people aged 80+ are institutionalised in, for example,
nursing homes and elderly homes.
In projecting future expenditure it is also important to know how “steep” the profiles are and whether
these “risk profiles” are stable or change over time. The risk profiles shown in Figure 2 are much
steeper than those used by the European Commission in its health care expenditure projections (see
European Commission 2015, Graph II.2.1):
As the risk profile for long-term care expenditure is much steeper than that for health
expenditure one can expect steeper profiles where the share of dependent individuals
in long-term care is greater. This is partly the case because the costs related to long-
term care are very high for institutionalised individuals, and the share of institutionalised
Class CZ DE LV LT HU NL SI FI SE(¹)
0 1.8 0.7 2.7 2.1 1.9 0.9 1.5 1.9 2.0
1-4 2.5 1.5 2.2 3.5 1.9 1.5 2.2 1.8 2.2
5-9 2.2 1.8 2.5 2.2 2.1 1.9 2.2 2.1 2.0
10-14 2.1 2.1 2.6 2.5 2.3 2.9 2.0 2.6 2.3
15-19 2.3 2.3 3.2 3.1 2.7 3.0 2.2 3.1 3.9
20-24 2.6 2.5 2.5 3.3 2.3 3.6 2.5 3.0 3.8
25-29 3.4 2.9 3.8 3.2 3.0 3.9 3.2 3.6 4.2
30-34 4.3 3.4 3.9 3.6 4.0 4.0 4.6 4.1 4.4
35-39 5.3 3.4 4.3 4.2 5.0 4.1 4.9 3.8 4.6
40-44 4.7 4.9 4.6 5.2 4.7 5.0 5.0 3.6 4.8
45-49 5.2 6.2 5.3 5.8 5.2 5.9 5.4 4.6 5.7
50-54 5.8 6.7 7.6 7.8 6.8 6.5 6.8 5.6 5.9
55-59 8.4 7.1 8.5 8.1 10.8 6.7 7.7 6.8 6.7
60-64 10.5 7.9 9.5 8.3 10.6 7.2 8.9 8.4 8.1
65-69 11.7 7.9 9.1 9.3 10.0 7.8 8.5 9.7 9.9
70-74 9.3 10.4 10.5 9.8 9.2 7.2 8.2 8.4 8.4
75-79 7.3 9.4 8.5 8.7 7.5 7.5 8.1 8.5 7.4
80-84 6.0 8.7 5.6 9.3 5.6 8.0 7.6 8.4 6.5
85-89 3.3 6.1 2.6 : 3.0 7.1 5.3 6.4 4.6
90-94 1.1 3.4 0.5 : 1.1 4.1 2.5 2.9 2.0
95+ 0.1 0.8 0.1 : 0.2 1.3 0.7 0.7 0.4
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
individuals increases sharply among persons aged over 80.
Figure 2: Health expenditure profiles per capita between 2012 and 2013
(country average per capita = 1)
0
2
4
6
8
10
0
Change of risk profiles between 2012 and 2013 One may expect only small changes of the risk profiles, when comparing consecutive years. Several
factors contribute to the change:
population changes (socio-demographic composition),
changes in expenditure (volume and prices),
other causes (e.g. ad hoc major health events such as epidemic outbreaks), and
changes in recording and coding methodology.
A one year period (2012 -2013) is certainly too short to expect major changes. It will clearly be
interesting to analyse longer periods to understand better the changes in age-related risk profiles.
For some countries time series are currently available which would allow such analysis.
Another important issue in the context of population effects on health expenditure is the spending on
health care treatment in the last years of life. Zweifel, Felder, Meier 1999 showed that the cost of
dying was significant during the number of quarters remaining until death while the age of the
persons was not. Therefore, a naïve estimation that does not control for proximity to death will
grossly overestimate the effect of population ageing on aggregate health care expenditure.
Sex-specific issues The patterns of age-related health care expenditure profiles significantly differ by sex. Broad causes
of disease for girls under 5 years are congenital abnormalities, preterm birth complications, lower
respiratory infections, neonatal encephalitis and sepsis, iron-deficiency anaemia, diarrhoeal diseases
and sudden infant death syndrome. Among girls aged 5–14 years, they are road injuries, asthma,
major depressive disorders and anxiety (see WHO Europe 2015).
In order to compare health expenditure between men and women, cost for pregnancy and
reproduction should be separated from other costs (See OECD 2008 p. 30). Therefore, countries
were asked to report specifically on data sources for reproductive health expenditures. However, the
separation of these costs is rather difficult on the level of the HPDS. The WHO ‘Guide to producing
reproductive health subaccounts within the national health accounts framework’ (2009) lists relevant
activities and conditions/diagnoses, which should be considered here (see also SHA 2011, Annex A,
p 390: HC.RI.3.1 Maternal and child health, family planning and counselling).
(%)
(¹) 2012 instead of 2013 (:) not available
Table 3 gives an overview of the distribution of health expenditures for women’s health by age
classes.
Table 4 gives an overview on the share of health expenditures devoted to the chapter “O00–O99
Pregnancy, childbirth and the puerperium”. Bulgaria, Greece, and Latvia report the highest spending
shares for pregnancy, while the Czech Republic, Hungary, Slovenia, and Germany have the lowest
shares.
In interpreting the results shown in Table 3 and Table 4 we must consider the coding practices of
health expenditures on pregnancy, childbirth and the puerperium by ICD-10 in the countries. For
example, the low value for the Czech Republic in Table 4 is not backed by a low expenditure ratio for
the age class 30-34 in Table 3. For this age class 30-34 Lithuania shows the lowest value, but, in
contrast, nearly the threefold expenditure share in Table 4. Care for physiologic pregnancy (including
delivery) would be largely covered by the Z30-Z39 codes, which could not be further elaborated from
the minimum dataset.
85-89 4.3 8.1 3.6 : 3.9 9.1 6.7 5.1
90-94 1.5 4.8 0.8 : 1.5 5.8 3.4 2.5
95+ 0.2 1.1 0.1 : 0.2 2.0 1.0 0.6
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Not allocated 0.3
Table 4: Health expenditures for women on ICD chapter O: Pregnancy, childbirth and the
puerperium in 2013
(¹) Part of pregnancy care is coded under the ICD10 chapter 21 and is therefore excluded from this table. (²) Also includes care for healthy infants Z32-Z39. It is not possible to separate these from O00-O99. (³) 2012 instead of 2013 (:) not available
Outlook This chapter has discussed variation in health expenditures by age and sex. Compilation of health
expenditures by age and sex will improve the data base for the analysis of the impact of
demographic changes on health expenditures. Almost all countries and European Institutions make
such investigations (see European Commission 2015).
Population ageing in EU Member States will continue to increase demands on health and long-term
care systems6 in the years ahead. It would be interesting to estimate CHE costs for 2020, 2030 and
2040 using the current Eurostat population forecast, utilizing the age cost-profile of 2013 and other
years. In addition we could calculate the percentage of expenditures for the age groups 65+ and
75+.This would show the demographic pressure of health care expenditure on total budget.
In this context an issue for further investigation is the connection between steepening health
expenditure profiles and expenditures at the end-of-life (see Gregersen, 2014). If there is a
substitution effect between different health care services, excluding other services could potentially
lead to biased results. It may be plausible that the profile steepens for inpatients, but the opposite
effect is observed in other health care services. Additional research should therefore take place in
other parts of the health care sector in order to confirm this pattern outside inpatient care.
The complex interplay of biological, behavioural, psychological and social protective and risk factors
contributing to health expenditures across the female and male lifespan also requires further
attention. Applying a life-course approach includes looking at women’s health needs beyond their
potential role as mothers. Women are living longer but have an increased risk of developing disease
and disability earlier in life. This is partly due to threats from non-communicable diseases and their
risk factors (see WHO Europe 2015).
6 In HEDIC LTC is limited to health; the social care components are almost certainly larger, and probably growing faster.
Countries O00–O99
Estonia : 4.9 : 14.7 : 1 434.3
Greece 3.4 7.2 51.5 50.3 6 001.8 5 854.2
Latvia 3.3 5.9 21.1 14.0 2 052.9 1 364.1
Lithuania 2.7 4.6 9.2 4.0 905.2 392.3
Hungary 1.6 3.1 21.3 13.0 2 374.8 1 447.1
Netherlands (²) 2.7 3.9 103.6 55.8 10 160.9 5 469.6
Austria : 3.8 : 36.3 : 3 871.9
Slovenia 1.8 5.5 34.7 29.7 3 386.9 2 895.5
Finland 2.4 4.6 41.9 32.1 3 919.0 3 002.1
Sweden (³) 2.2 4.9 41.7 35.7 3 521.3 3 011.1
Chapter summary Interest in expenditures by age and sex has increased, with increasing consideration of the
impacts of population ageing for health care systems and health care financing.
Analysis of the HPDS shows that health expenditures vary by age among countries in 2013.
The age–related expenditure profiles for 2012 and 2013 are compared for five countries:
Czech Republic, Germany, Finland, Lithuania, and Slovenia. The changes are small
because they are for two consecutive years, but may be due to changes in the socio-
demographic composition of the population; the evolution of volume and prices; and ad hoc
major health events such as epidemics.
In this study a preliminary investigation of the determinants of sex differences has been
made.
In order to compare expenditure on men and women, it is very important to separate the
cost for pregnancy and reproduction (the latter for both men and women), from other costs.
The same applies to sex-specific diseases such as ovarian and prostate cancer.
Data from the HPDS are presented on expenditure by age for women’s health by ICD10 for
2013; and on expenditure on pregnancy, childbirth and the puerperium by ICD10 as
percentage of Allocated Current Health Expenditure (ACHE), as percentage of Allocated
Inpatient Health Expenditures (AIHE), per capita and per 1,000 live births.
It is necessary to consolidate the compilations and to decompose profiles across financing
and provision as well as over time.
We need to investigate further the differences and dynamics of profiles, which among other
things will need longer time series of data.
Ongoing analysis of long-term care developments is needed to understand differences in
risk profiles.
Further work is needed to develop standardised indicators.
Outlook: Compilation of health expenditures by age and sex will improve the data base for
the analysis of the impact of demographic changes on health expenditures, in a context of
increasing demand for health and social care as Europe’s population gets older.
References Astolfi, R., Lorenzni, L., Oderkirk, J. (2012), A comparative analysis of health forecasting methods,
OECD Health Working Paper No. 59, Paris.
European Commission (2014), The 2015 Ageing Report: Underlying Assumptions and Projection
Methodologies, Joint Report prepared by the European Commission (DG ECFIN) and the Economic
Policy Committee (AWG), Economic and Financial Affairs, EUROPEAN ECONOMY 8|2014.
European Commission (2015), The 2015 Ageing Report: Economic and budgetary projections for the
28 EU Member States (2013-2060), Economic and Financial Affairs, EUROPEAN ECONOMY
3|2015.
Finkenstädt, V., Niehaus, F. (2015), Länderrankings auf Basis der OECD-Gesundheitsdaten – Eine
Analyse der methodischen Probleme" (Country rankings based on OECD Health Data - An analysis
of the methodological problems), in: Monitor Versorgungsforschung (MVF) 04/15, S. 44-50.
Gregersen (2014), Ageing, mortality and health care expenditures: The case of Norwegian hospitals
and ambulances. Dissertation, Faculty of Medicine, University of Oslo, No. 1765.
OECD (2008), Estimating Expenditure by Disease, Age and Gender under the System of Health
Accounts (SHA) Framework.
OECD, Eurostat, WHO (2011), A System of Health Accounts 2011, OECD Publishing, Paris.
WHO (2009), Guide to producing reproductive health subaccounts within the national health
accounts framework, Geneva. http://www.who.int/nha/docs/guide_to_rh/en/index.html.
WHO Europe (2015), Beyond the mortality advantage: Investigating women’s health in Europe,
Copenhagen.
Zweifel, P., Felder, S., Meier, M. (1999), Ageing of population and health care expenditure: a red
herring? Health economics, 8, 485-496.
Grouping of diseases HEDIC uses the International Classification of Diseases (ICD) of the World Health Organisation
(WHO) in the attribution of health care expenditure according to disease. The sheer size of the ICD
classification, which contains many thousands of diseases, requires them to be grouped. Based on
the country studies of earlier projects, HEDIC groups them into chapters, as a first step.
Former international comparisons show rather similar shares of total health care spending by
chapters of diseases among developed countries (see Heijink et al 2008, Slobbe et al 2009). Our
initial comparisons among HEDIC countries do not confirm this result. One major reason may be that
those studies compared a limited set of providers of Western European countries but excluded
countries in Eastern Europe. Furthermore, differences might be explained by differences in exclusion
or inclusion of specific functions or providers. Below we look more closely at these variations. After
showing the deviations from the standard structure, we analyse three major disease categories.
Circulatory Diseases: Expenditure is highest for circulatory diseases in most countries. We
suppose that the growth of this expenditure is stagnating because treatment is getting cheaper. As a
consequence, the share of health expenditure devoted to circulatory diseases is diminishing
compared to previous years.
Neoplasms: Expenditures for neoplasms are increasing because European populations are ageing.
Unit costs of treatment are sometimes very expensive. As a consequence the share of health
expenditures devoted to neoplasms is increasing. This is only part of the story. Improved survival
rates and longer treatment periods contribute also to this rise.
Mental Diseases: Expenditure for mental disease is also increasing, partly as result of population
ageing, partly with rising living standards.
In order to separate the ageing effect from other growth determinants, it is necessary that a
multidimensional HEDIC data set is compiled, such as those for the Netherlands and Finland.
Variation of profiles among countries This subsection describes the variation of health expenditures by ICD Chapters among HEDIC
countries.
Various factors may explain international variations in the share of health expenditures devoted to
different disease chapters, including: 1) differences in prevalence / incidence and in demand for
treatment; 2) differences in access to health care services and the local supply of technology; and 3)
differences in coding and reporting practices.
Differences in reporting practices influence substantially the presentation of data for disease and age
related health expenditures. For example, in the case of Bulgaria, the only reliable source of data for
5 Disease related expenditure profiles
inpatient and some providers of outpatient care is the National Health Insurance Fund. Estimations
for disease disaggregated data for Central and Local Government and out-of- pocket expenditures of
households are only possible for some classes of diseases.
On average, health expenditures for ICD chapter 9 “Diseases of the Circulatory system” is the most
important category in all countries, including about one sixth of current health expenditures (see
Table 5). Other major expenditure categories are chapter 2 “Neoplasms”, chapter 5 “Mental and
behavioural disorders”, chapter 11 “Diseases of the digestive system”, and chapter 13 “Diseases of
the musculoskeletal system and connective tissue”.
(%)
Ann.: The percentages are standardized on the sum of the allocated health expenditures in each country. (¹) structure refers to total inpatient and outpatient expenditures for 2013. (²) expenditures for GPs and households-financed care were not completely allocated and are therefore
not fully included. (³) 2012 instead of 2013 (:) not available
When interpreting Table 5, we should also bear in mind expenditure which is not reported. The not
allocated part of current health expenditures varies between 32.1 % in Bulgaria and 0.8 % in
Lithuania, and 2.1% in Germany.
Expenditures for chapter 11 “Diseases of the digestive system are particularly high due to dental
expenses in Bulgaria, Sweden and Germany.
Chapter 21 “Factors influencing health status and contact with health services” contains expenditures
for interventions other than for a disease, injury or external cause. It is designed to include check-
ups, screening, normal reproduction, etc. The coding practices for chapter 21 seem to be differently
applied between countries and not directly comparable.
Some countries used the ICD 10 chapter 22 “Codes for special purposes”, which contains new
diseases of uncertain aetiology or emergency use and encounters with resistance to antimicrobial
and antineoplastic drugs. The codes of chapter 22 are not always accessible in electronic systems.
ICD 10 Description BG(¹) CZ(²) DE EL LV LT HU NL SI FI SE(³)
I Infectious 2.0 2.3 1.9 1.5 3.0 3.5 2.4 1.4 2.2 2.1 2.0
II Neoplasms 8.4 10.0 8.4 12.5 8.0 9.7 13.1 7.7 9.3 11.9 7.4
III Blood 0.6 1.1 0.8 1.9 1.1 1.2 2.0 0.7 1.1 1.0 0.7
IV Endocrine 2.9 5.8 5.0 9.2 4.0 4.5 7.9 3.8 3.0 5.1 3.4
V Mental 2.2 5.3 11.1 7.4 10.7 6.6 6.8 24.8 8.3 11.6 9.8
VI Nervous 2.3 4.0 3.5 2.9 4.2 4.1 4.7 8.3 4.1 5.7 2.6
VII Eye 3.0 3.5 1.8 2.4 5.4 3.8 2.1 : 4.4 1.8 1.9
VIII Ear 1.1 0.6 1.3 0.4 2.3 1.2 1.1 : 0.9 0.9 1.1
IX Circulatory 22.5 17.2 13.8 16.9 19.2 23.5 16.6 12.9 12.8 15.3 10.4
X Respiratory 7.4 6.7 6.4 5.5 6.8 8.2 7.2 4.8 5.4 6.2 4.8
XI Digestive 19.4 11.6 14.0 10.4 8.5 9.5 7.0 9.0 9.8 8.8 15.8
XII Skin 1.6 1.5 1.4 0.6 1.4 2.2 1.8 1.6 1.6 1.4 1.9
XIII Musculoskeletal 5.0 7.5 11.7 7.5 7.2 6.5 8.5 8.3 7.9 7.3 8.1
XIV Genitourinary 8.1 6.4 4.2 6.5 5.2 4.4 4.7 4.1 5.4 4.0 3.4
XV Pregnancy 3.1 1.1 1.8 3.4 3.3 2.7 1.6 2.7 1.8 2.4 2.2
XVI Perinatal 0.4 0.9 0.3 0.9 0.7 1.1 0.7 0.2 0.5 1.1 1.0
XVII Congenital 0.6 0.4 0.4 0.3 0.6 1.0 0.5 0.4 0.8 0.9 0.8
XVIII Symptoms 0.6 3.8 5.1 4.2 0.2 0.8 3.0 5.8 4.5 3.5 6.2
XIX Injury : 4.3 4.4 2.9 6.5 5.3 3.8 3.6 6.8 6.1 6.8
XX External 2.6 0.1 : 0.2 0.1 : 0.2 : 0.0 0.0 0.0
XXI Factors 6.1 6.0 2.7 2.6 1.9 0.3 4.3 : 9.5 2.8 9.7
XXII Special 0.0 0.0 0.0 0.0 2.6 0.8 2.1 : 0.0 0.0 0.0
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Not allocated 32.1 10.0 2.1 11.0 2.6 0.8 2.1 15.1 : : 12.9
Variation of expenditures by disease between 2012 and 2013 In recent years, and in particular after the economic crisis in 2008, European countries have
implemented or strengthened a number of cost-containment policies (see OECD 2015). HEDIC
provides information which helps to assess the impacts of these policies on resource allocation by
disease.
Figure 3: Deviation of growth of health expenditures by disease from national average,
2012/2013
(:) not available
The analysis of health expenditures by disease over time can give further insight into the
consequences of cost containment policies. Economic, fiscal and health policy reacted rather
differently on fiscal deficits. In Greece for example, the economic crisis led to a massive reduction of
current health expenditures. But, impacts on the treatment of disease were not uniform. In Greece,
some diseases areas lost more resources than others. On average health expenditures fell 9.6
percent in the period 2012-2013. Figure 3 shows the deviation of growth in each disease area as
compared to the national average growth rate.7 Especially, expenditures for diseases of the skin and
the subcutaneous tissue, diseases of the musculoskeletal system and connective tissue, and for
certain conditions originating in the perinatal period were cut. Cost containment in Bulgaria focused
on certain conditions originating in the perinatal period, but also on certain infectious and parasitic
diseases as well as factors influencing health status and contact with health services. At this stage of
the investigation it is not possible to assess the public health consequences of these policies. An
analysis based on several years would certainly be useful.
7 Please note that the national growth rate for the health care activity j varies across countries. The deviation wdj – wj between the
growth of the disease specific expenditures wdj and the growth on average wj allows comparing the structural change of the disease expenditure across countries.
ICD 10 Description BG CZ DE EL SI FI
I Infectious -12.4 5.4 1.9 -0.3 4.6 -1.3
II Neoplasms 3.2 3.8 -0.3 -2.5 9.2 8.0
III Blood -0.5 6.9 2.3 4.0 4.8 2.9
IV Endocrine 1.2 -3.0 -0.3 10.4 0.7 4.5
V Mental 33.2 -3.9 2.2 1.5 -2.2 -1.1
VI Nervous 3.0 1.8 0.3 -1.5 -1.1 -3.7
VII Eye 3.4 -3.3 -0.7 2.3 1.3 3.4
VIII Ear 2.4 3.7 -2.5 4.5 8.1 5.2
IX Circulatory 2.4 -2.6 -3.4 1.4 -1.4 -4.0
X Respiratory -0.1 2.6 6.0 1.9 2.2 -4.2
XI Digestive -1.3 1.2 0.4 -2.8 -4.4 5.2
XII Skin 0.0 -2.7 1.2 -19.8 6.9 3.0
XIII Musculoskeletal 8.4 -4.1 0.6 -6.2 12.4 4.4
XIV Genitourinary -0.4 -1.6 -4.9 -2.5 -10.4 -3.7
XV Pregnancy -2.5 1.8 1.1 -0.4 4.3 13.1
XVI Perinatal -17.7 5.5 0.4 -7.2 -24.0 13.0
XVII Congenital 0.6 -0.9 3.2 0.9 -5.8 16.1
XVIII Symptoms 1.3 4.4 -0.6 0.7 8.1 1.9
XIX Injury -0.2 1.6 1.0 -1.6 -10.0 -0.3
XX External -1.2 7.4 5.6 -1.4 -9.3 :
XXI Factors -5.6 6.0 2.0 2.9 -1.2 :
XXII Special : : : : : :
The analysis of health expenditures by disease over time can give further insight into the
consequences of cost containment policies. The data presented in Figure 3 show that, in Greece for
example, the economic crisis led to a massive reduction of current health expenditures. Especially,
expenditures for diseases of the skin and the subcutaneous tissue, diseases of the musculoskeletal
system and connective tissue, and for certain conditions originating in the perinatal period were cut.
Cost containment in Bulgaria focused on certain conditions originating in the perinatal period, but
also on certain infectious and parasitic diseases as well as factors influencing health status and
contact with health services. At this stage of the investigation it is not possible to assess the public
health consequences of these policies. An analysis based on several years would certainly be useful.
Circulatory diseases Cardiovascular Diseases (CVD), defined by the ICD-10 codes I00-I99, cover a range of illnesses
related to the circulatory system, including heart attack and cerebrovascular diseases such as stroke.
The European Heart Network (EHN) has published several studies of the cost of CVD (see for
example Nichols et al, 2012). The methodology applied differs from that used in HEDIC for five main
reasons: (1) The EHN approach focuses only on one group of diseases; (2) Costs are derived using
aggregated data on morbidity, mortality, hospital admissions, disease related costs, and other health
related indicators from various national and international sources, partly at different times, but
updated to the year 2009; (3) The boundary of CVD health care services is narrower than SHA,
excluding preventive activities, physiotherapy, long-term nursing care, medical devices, and
administration; (4) Private spending was often estimated by using the total proportion of private
spending on health care. (5) Age and sex specific expenditures were not compiled.
The EHN estimates the CVD cost to the health care systems of the EU at just over €106 billion in
2009. This represents a cost per capita of €212 per annum, around 9% of the total health care
expenditure across the EU. The cost of inpatient hospital care for CVD patients accounted for about
49% of these costs, and that of drugs for about 29%. The amount spent on health care for people
with CVD varies widely across the EU. Cost per capita varied ten-fold in 2009, from €37 in Romania
to €374 in Germany expressed in exchange rates8. Percentage of total health care expenditure on
CVD varied from 4% in Luxembourg to 17% in Estonia, Latvia and Poland (see Nichols, Townsend,
Luengo-Fernandez et al. 2012).
As a consequence of the different approaches used by HEDIC and EHN one might expect that the
HEDIC analysis would show higher expenditures for CVD both in total, and as share of current health
expenditures. In fact, Table 6 shows for all countries that the HEDIC approach leads to higher
expenditure estimates for circulatory diseases than the EHN study.
8 The EHN study used the exchange rate on the last day of 2009 (see Nichols et al, 2012).
(%)
Ann.: The share refers to the sum of the allocated expenditure across all ICD10 chapters. (¹) 2012 instead of 2013
A one year comparison 2012/13 is certainly not enough to reach conclusions about expenditure
trends by disease, partly due to uncertainty in measuring this variation. Compare for example the
Netherlands where the share stays about the same between 2003 and 2011. The most likely
explanation is that the decreasing incidence and possible lower treatment costs (of statins for
example) are largely offset by an increase in life-expectancy for patients, combined with ageing, and
that prevalence stays about the same.
Neoplasms Here we compare HEDIC data with estimates for cancer costs based on data published in the Lancet
Oncology (Luengo-Fernandez et al, 2013). The researchers from Oxford University and King's
College London (OUKCL) estimated direct and indirect costs of diseases of ICD chapter C00-C97 in
the year 2009 for 27 EU MS. They used the same macro approach for this group of diseases as in
the case of CVD (see section 0). Costs associated with breast (C50), colorectal (C18–21), lung
(C33–34), and prostate (C61) cancers were estimated separately. The results revealed substantial
disparities between different countries in the EU in spending on healthcare and drugs for cancer. (It
should be noted that for many European countries there were large gaps in the data in this study.
These were partially filled by utilizing German and Dutch data. which were available in much more
detail than in other countries.)
In the OUKCL study, Luxembourg and Germany spent the most on healthcare for cancer per person,
with Bulgaria spending the least. The researchers conclude that these results show wide differences
between countries, the reasons for which need further investigation. However, these data contribute
to public health and policy intelligence, which is required to deliver affordable cancer care systems
and inform effective public research funds allocation.
Countries
HEDIC
Bulgaria 22.5 13.0 57.7
Germany 13.8 11.0 79.9
Greece 16.9 11.0 65.2
Latvia 19.2 17.0 88.4
Lithuania 23.5 12.0 51.0
Hungary 16.6 14.0 84.1
Netherlands 12.9 8.0 61.8
Slovenia 12.8 8.0 62.5
Finland 15.3 12.0 78.4
(%)
Ann.: The share refers to the sum of the allocated expenditure across all ICD10 chapters. (¹) 2012 instead of 2013
Table 7 shows for all countries that the HEDIC approach leads to higher expenditure estimates
devoted to cancer diseases than the Luengo-Fernandez et al 2013 study: They have underestimated
the cost of cancer as compared to HEDIC. In addition to the use of a narrower boundary, one reason
seems t

Date post:	18-Dec-2021
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

HEALTH EXPENDITURES 2016 edition BY DISEASES AND ...

Documents